This is an R21 application to determine the prognostic value of assays of genomic instability in human colon cancer. It focuses on the fact that instability, though determined by the presence of mutations in DNA mismatch repair genes, is a complex phenotype exhibiting differences in instability at different loci (eg oligo A versus microsatellite) and variability in quantitative extent of instability at a given locus. Thus, such variable penetrance of the mutations would have an impact on clinical utility of these markers. Further, clear demonstration of variable penetrance would raise the possibility that environment (eg diet), the other major risk factor for colonic cancer, can affect the instability phenotype. DNA already isolated from over 1300 tissue samples representing more than 700 patients on 2 phase III studies for adjuvant therapy for Duke' B and C tumors carried out by the Eastern Cooperative Oncology Group (ECOG) will be utilized to: 1) determine the prognostic value of genomic instability on recurrence and survival; 2) determine the relative prognostic value of instability at oligo A versus microsatellite sequences; 3) determine the prognostic value of a novel measurement of quantitative extent of genomic instability; 40 determine whether low levels of instability can be detected in normal mucosa of patients with Dukes' B and C tumors. Of particular note, one of the phase III studies which has supplied material contains a surgery only arm. Since adjuvant therapy is now recommended for all resectable Dukes'B and C patients, this represents the last opportunity to study the relationship of genetic alterations to the natural history of the disease. The loci to be investigated for instability include; oligo A sequences in two cDNA sequences which we have identified as overexpressed in flat colonic mucosa at risk for colon tumor development; a CA microsatellite and an oligo A sequence in the 3; transcribed, but untranslated region, of the cyclin D1 gene; an oligo A in the 4th exon of the PCNA gene; a CA microsatellite in the RB gene; and a CA microsatellite upstream of the apoD gene. Statistical analysis will be carried out through an ongoing collaboration with ECOG Biostatistics at the Dana-Farber, which maintains a complete clinical data base on these patients. Moreover, novel statistical methods will explore the relationship between genetic instability and risk of relapse.